Social cognition in the we-modeMattia Gallotti1 and Chris D. Frith2,3,4

1 Jean Nicod Institute, Ecole Normale Supe rieure, 29 rue d’Ulm, 75005 Paris, France2 University College London, 12 Queen Square, London WC1N 3BG, UK3 All Souls College, 27 High Street, Oxford OX1 4AL, UK4 Interacting Minds Centre, Aarhus University, 4 Jens Chr. Skous Vej, 8000 Aarhus C, Denmark

According to many philosophers and scientists, humansociality is explained by the unique capacity to share themental states of others. Shared intentionality has beenwidely debated in the past two decades in ways that alsoenlighten the current ‘interactive turn’ in social cogni-tion. In this article, we examine the function and signifi-cance for interacting agents of sharing minds in anirreducibly collective mode called the ‘we-mode’. Thisfirst-person plural perspective captures the viewpoint ofindividuals engaged in social interactions and thusexpands each individual’s potential for social under-standing and action. This proposal shows that a non-reductionist, interaction-based approach can be devel-oped that nevertheless resists recent suggestions con-cerning the constitutive role of interaction for socialcognition.

Individualism in social cognitionTraditionally, reflections about the nature and develop-ment of mindreading have been central to inquiries intosocial cognition (see Glossary). On the classic view, mind-reading is the capacity of people to ascribe mental states toothers [1,2]. Such mindreading is viewed as the outcome ofcognitive processing that occurs in an individual’s mind inabstraction from, and as a precondition for, interactionwith others [3]. This is mindreading achieved throughobservation. Recently, this view has met with criticismfrom disciplines concerned with the problem of socialinteractions, whether these be low-level spontaneous epi-sodes of coordination or complex instances of planned jointaction [4–6]. The argument of the current ‘interactive turn’in social cognitive research is that, when interacting,agents appear to have access to more information aboutthe behaviour of their partners than they would as mereobservers in a disembodied social context. For example,according to the emerging literature on the role of ‘second-person’ engagement in social cognition, individuals en-gaged in real-time social interaction can attain a greaterunderstanding of the goals of others and can use thisevidence to ascribe higher-order mental states. This hasimportant consequences for the way social cognition istheorized about and investigated empirically [7,8].

However, despite widespread agreement on the impor-tance of studying real-time interactions, the questionremains as to how individuals involved in a joint action

Opinion

Glossary

Cognitivism: a variant of individualism holding that mental states ought to be

individuated so as to supervene on the internal (i.e., brain) operations of the

individuals having those states.

Collective intentionality: the complex of representational features character-

izing the mental events and episodes that are the proximate causes of joint

action. Theories of collective intentionality fall by and large in two families

depending on whether those features are viewed as attributes of shared

states of affairs, such as plans of action, or of the cognition of interacting

agents, such as the mode in which they represent aspects of the action

scene.

Co-representation: the capacity of people to keep track of their own, as well as of

actual or potential interacting partners’ actions, and to monitor performance in a

social setting. Theories of co-representations differ depending on how they

characterize the content of co-representations, be it the other person’s task

contribution to a joint task or aspects of the other person’s task (like when it is

the other person’s turn to react).

Enactivism: the approach to social cognition according to which interpersonal

understanding and action are constituted by relational dynamics in which

autonomous systems co-regulate their coupling as part of the physical and

social environment.

Individualism: the general view that all sorts of complex behaviours entail

properties of the individual as distinct from other levels of functional

organization. The entailment relation lends itself to epistemic, semantic,

and ontological interpretations. A theory of social cognition is (anti) indi-

vidualistic if it implies that understanding of other minds is (not) metaphys-

ically determined in abstraction from the individual’s social and physical

environment.

Interactionism: a family of embodied, embedded, enactivist, and extended

approaches to cognition, holding the view that explanations of social cog-

nition do not necessarily involve reference to the internal operations of the

brain.

Irreducible collective behaviour: the claim that joint action cannot be fully

explained in terms of the assumptions of individual choice presupposed by

causal theories of (joint) action. The we-mode theory of collective intentionality

aims to articulate such a claim by assuming that one’s potential for social

understanding and action is enlarged by cognizing in a dedicated collective

mode.

Joint action: any form of interaction involving at least two agents that is made

fully intelligible by reference to representational features accessed by the

subject in the first-person plural.

Mode: the property of mental representation that captures the subject’s per-

spective or attitude on the intentional object. Intentional mental states are

representations characterized by the object they represent (are ‘about’) and

by the aspectual shape in which the intentional object appears to the subject.

The mode of a mental representation is an additional feature of mental content

specifying whether action predicated over individuals is represented as some-

thing that each person individually intends to pursue or as something to be

pursued together with others (joint action). Our use of the term ‘mode’ is not

meant to capture the difference between types of intentional mental states, or

‘intentional modes’, such as believing, desiring, intending, imagining, and so

forth.

Social cognition: the processes that sustain people’s understanding of, and

interaction with, others. Theories of social cognition reflect distinct ways of

thinking about the nature of human cognition in terms of processes such as

simulation or theorizing (or both) and/or by relational processes between an

organism and its environment.

Team reasoning theory: among theories of collective intentionality that empha-

size the role of cognitive features, this is the view that joint action is underpinned

by reasoning in accordance with inferential schemas that represent action as

directed to the best outcome for the group (team). The process whereby a

situation is conceptualized from the point of view of the group is described as

involving ‘preference’ or ‘agency’ transformation.1364-6613/$ – see front matter

� 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.tics.2013.02.002

Corresponding author: Frith, C.D. (c.frith@ucl.ac.uk)

160 Trends in Cognitive Sciences April 2013, Vol. 17, No. 4

can have their abilities augmented by acting together withothers. An influential answer is represented by ‘interac-tionism’, a family of views motivated by dissatisfactionwith the individualistic tendency of psychological researchto reduce group psychology to the psychological features ofsingle agents. Generally, interactionists hold that, whenagents are poised to interact, they achieve interpersonalawareness through a ‘meeting’ of minds rather than anendlessly recursive exercise of mindreading [9]. Whenexpressed in these terms, the importance of interactionfor interpersonal understanding and joint action forges aninteresting link between current controversies about thenature of social cognition and philosophical discussions ofjoint action. By and large, philosophers agree that agentsmust have their mental states qua intentions ‘shared’ foran action to be joint [10], yet they disagree about theconditions that bring about the relevant sharing of minds.For some theorists of shared intentionality, agents mustsee their actions as directed to something that they aregoing to pursue together (as a ‘we’) if action is to count asjoint [11,12]. This sense of ‘we-ness’ is a striking feature ofthe psychology of collective behaviour, hence the view thatinteracting agents have their minds shared by cognizing inan irreducibly collective mode of cognition called the we-mode.

In this article, we propose a theory of the we-mode thatcaptures the role of interaction for expanding the social-cognitive resources of individuals. Our proposal is thatindividuals engaged in joint action have a broader under-standing of the behaviour of their partners, and thus ofoptions available for action, by representing aspects of theinteractive scene in the we-mode (Box 1). With this pro-posal, we aim to offer a balanced response to the demandfor an interaction-based approach to social cognition: we

Box 1. Groups with minds vs minds in the we-mode

Since its appearance, collective intentionality has been given a

prominent role in accounts of the foundations of human sociality

because it addresses the question of how individual agents come to

intend and pursue things together [51]. Yet, not every episode of

interaction between at least two persons can be classified as such.

Actions involving more than one agent acting on their own are

merely accidentally, not intentionally, collective.

One example will help to illustrate the difference. The sentence

‘Renzo Piano and Richard Rogers designed the Pompidou Centre’

expresses the idea that each architect made his own contribution to

the final creation. However, it also allows a collective reading of the

sentence: what they did – they did it jointly. On this reading, the

project was a truly collective outcome resulting from the two

architects acting as a group, as opposed to the meaning of ‘Piano,

like Rogers, designed parts of the Pompidou Centre’, which

suggests that action predicates are distributed over the individuals.

This is to say that, when two or more agents come together and act

as a group in achieving a collective goal intentionally, the statement

that they do something together can be read as suggesting that no

member of the group does it ‘on her own’ [52].

Prima facie two responses are conceivable to the question of how

to conceptualize, and account for, the specific attitude that under-

pins collective intentional behaviour. The first response charac-

terizes the irreducibility of group behaviour as a feature of the

bearer holding the relevant attitude, namely the group (rather than

the individuals forming the group). The possibility to ascribe a ‘we-

as-a-group’ attitude implies that there is a plural subject, that is, a

minded group, to which the attitude can be ascribed.

The second response focuses on features of the individuals

forming the relevant group [53]. In contrast to the former response,

in the early days of collective intentionality theory, philosophers

started to investigate individual-level features in terms of the type of

attitude that persons display when they intend and do something

with others, focusing on the way in which each represents the

intentional structure of a joint endeavour. The intuition that

behaviour can be guided by mental states that are collective in that

they are accessed in a first-person plural mode led to the emergence

of we-mode accounts of collective intentionality [11,12].

Opinion Trends in Cognitive Sciences April 2013, Vol. 17, No. 4

share the concern of interactionists regarding the exces-sively reductionist nature of classic mindreading theories,but we also draw on fundamental assumptions about therole of the individual in addressing the questions of social-ity more generally. In order to develop philosophicalinsights concerning the we-mode into a scientifically plau-sible model, we bring together various strands of researchon joint action in the cognitive sciences and the neuros-ciences.

Interactionist social cognitionAn influential attempt to capture the role of interaction forsocial cognition is represented by ‘interactionism’ [9], afamily of views unified by opposition to various forms ofindividualism, such as cognitivism. According to interac-tionists, the theory-of-mind research tradition has con-fronted the problem of social cognition in anindividualistic fashion, being concerned only with thequestion of what the individual brings to interaction. Socialunderstanding and action have thus been characteristical-ly depicted as the output of cognitive processing takingplace in individuals [13]. In contrast to this view, mostinteractionists embrace the enactivist position that the lifeand cognition of agents are regulated by their dynamicencounters with the physical and social environment [14].These encounters are not reducible to attributes of theindividual mind, because it is the interactive unit that

behaves in a certain way, and the cause of this behaviour iscaptured by the collective dynamics themselves. Interac-tion thus offers not only a context for social cognition but,most emphatically, it can constitute individuals’ socialcognitive resources in a way that need not be mediatedby changes intrinsic to the individual [5].

Although it is crucial to stress the enabling role of theenvironment for interpersonal understanding, in general,we believe that claims about its allegedly constitutive rolemiss the point of the interactive turn in social cognitiveresearch. To see why, note that enactivism implies twoclaims about (social) cognition. One is that cognitive activ-ity consists predominantly in making sense of things in theworld, where sense-making is the relational process be-tween an organism and its environment that transformsthe world into a place of meaning and value [15]. When theenvironment is social, sense-making occurs in a participa-tory manner [14]. The other claim concerns the scope of‘participatory sense-making’. To say that meanings emergeand are continuously negotiated by virtue of individuals’interactive coupling with the social world amounts tosaying nothing more than that interaction dynamics defineand constrain the content of individual minds [16,17].Although the former claim, namely, that features of thephysical and social environment shape cognition, has beenwidely scrutinized, for example, in the extended-mind [18]

161

Opinion Trends in Cognitive Sciences April 2013, Vol. 17, No. 4

and mechanistic literature [19], somewhat less attentionhas been devoted to assessing the scope of participatorysense-making, that is, the extent to which enactivismsucceeds in capturing the role of interaction, not just forcognitive processes in general, but for social cognition morenarrowly.

The idea of participatory sense-making is that centralaspects of (individual) cognitive performance, notablymeaning-formation, are inherently relational. It is byinteracting with physical and social objects, includingthe minds of other people, that the world becomes mean-ingful to individuals. How does such a view answer thequestion of the role of interaction for social cognition? Thatis, in what sense do interacting agents get to know moreabout the behaviour of others in the context of interactionthan they do from a detached point of view? One answer isthat cognition is necessarily modulated by interaction withothers, because meanings are acquired and shared throughinteractive practices – a point that most enactivists andcognitivists alike would be happy to accept. However, thereis a crucial difference between making sense of the worldwith others and making sense of others as part of the worldpeople live in [20]. The interactive turn in social cognitionimplies that, when people engage with others, there mightbe something unique to the way in which they make senseof others’ minds that is not reducible to the kind of cogni-tive processing at work when understanding other worldlyofferings. We believe that enactivists miss this differenceinsofar as participatory sense-making is presented as yetanother formulation of the claim that meanings are createdthrough the interaction among people in general. Indeed,this claim stands as a fruitful reminder for cognitivists thatthe intentionality of the mind – how things in the world arerepresented as having certain meanings – cannot be caus-ally reduced to a series of processes going on in the head ofindividuals. Nonetheless, it should be clear that the social,that is, embedded, character of cognition is not underquestion in this paper. In the present context, we need atheory of the role of interaction for social cognition, namely,a theory that explains how novel routes to knowledge ofother minds become available to individuals who arepoised to interact, rather than a refined version ofthe argument for the social foundations of (individual)cognition.

Shared intentionality and the interactive turnWe agree with interactionists that the tendency of psycho-logical research to focus on the isolated observer is unsat-isfactory [21]. This tendency implies that the relationshipbetween mind and society can be explained simply bygeneralising claims about cognition within the individualto social cognition. In fact, agents engaged in a socialactivity achieve a result that is not reducible, at theappropriate level of description, to the sum of their singlecontributions. Indeed, when they act together in groups,individuals have access to information about the inten-tions, reasons, and emotions of their interacting partnersthat opens up novel possibilities for action unavailable toisolated observers.

For this very reason, our concern is not with the irre-ducibility of group to individual behaviour which, despite

162

disagreements, seems to unify advocates of the currentinteractive turn. We are rather concerned with the cogni-tive processes underlying group behaviour. This motiva-tion becomes especially vivid in episodes of sociality whereagents interact in a manner which seems rational for thegroup, but which cannot be reconstructed as rational forthe individual given traditional assumptions of how indi-viduals reason in social interactions. On these assump-tions, there are features of group behaviour that cannot befully explained by reference to mechanisms of individualcognition and agency [22]. Furthermore, sometimes indi-viduals appear to be thinking and acting in ways that areeven detrimental to the pursuit of their own benefit [23–26]. How, then, can a non-reductionist, interaction-based,approach to social understanding and action be accommo-dated in a theory that purports to avoid the most radicalclaims of interactionism?

We suggest that the problem with individualisticaccounts of social cognition stems from the underlyingmodel of how people cognize in the context of interaction.The spectatorial view of social cognition accounts for thebehaviour of interacting agents from the perspective of anobserver qua theorist, who represents the decision problemfaced by the agents as it appears to him. Yet, engaging ininteraction changes the way in which interacting agentsunderstand the problem, in that contextual features of theinteractive scene prime representations that are not avail-able to the isolated ‘theorist’ [27,28]. According to philoso-phers, if action is to count as truly joint action, it is notsufficient that individuals pursuing a collective goal eachindividually intend to contribute [10]. Sociality is not justphysical co-presence: it involves some actual or potentialunderstanding of aspects of the interactive scene as sharedby the participants in a joint action [29,30]. To illustratethe idea, imagine that Mary and John come across a friendin difficulty and offer help. It seems plausible that what-ever they mean to do in order to bring about joint assis-tance, they do it together. In other words, it is becauseMary and John see each other as being part of the same‘group’ that Mary understands John doing his part, andherself doing hers, as contributions to something that theyare doing together, rather than just as the result of indi-vidual tasks undertaken simultaneously. More generally,when people join forces and act as a group, there is a sensein which the fact that they do something together impliesthat no member of the group does it ‘on her own’. This senseof ‘we-ness’ is a striking feature of the psychology ofcollective intentional behaviour, hence the view that jointaction involves shared or collective or ‘we-intentions’ [31].

Few would doubt that the propensity to share mentalstates is a basic feature of humanity – perhaps the distin-guishing developmental trait of human cognition in theanimal kingdom [32]. However, there is much disagree-ment about the conditions under which people’s mentalstates are said to be shared and which method is mostappropriate to examine the irreducibility of shared mindsto attributes of the individual mind [33,34]. In the presentdiscussion, we will focus on one philosophical view ofshared intentionality, which characterizes the processesthat form and sustain shared minds in terms of an irre-ducibly collective psychological attitude or ‘mode’ [11,12].

Opinion Trends in Cognitive Sciences April 2013, Vol. 17, No. 4

We shall refer to these processes as we-mode processes(first-person plural) and present them as parts and opera-tions of the mechanism responsible for the ‘meeting’ ofminds that is essential to joint action. One major reason fortaking this route is dissatisfaction with the assumptionthat interactions are always guided by representations inthe head of agents representing states of affairs, includingothers’ minds, from the perspective of the thinking andexperiencing subject ‘I’. However, before we show how thisapproach offers a solution to the question of how individu-als have their potential for social understanding and actionexpanded in the context of interaction, we need to specifywhat a we-mode process is and how it works.

Cognizing in the we-modeThe central idea of the we-mode is that interacting agentsshare their minds by representing their contributions tothe joint action as contributions to something that they aregoing to pursue together, as a ‘we’. On such a view, cogniz-ing in the we-mode does not entail that individuals havemental representations of their task with the same, orsimilar, content, or a specific representation of ‘together-ness’, or ‘we-ness’. To represent things in the we-mode is for

CongruentGroup observed

IncongruentIndividual observed

C

C

P

P

TRENDS in Cognitive Sciences

Figure 1. We-representations and me-representations. Background: according to

prior research [54], individuals tend to co-represent the action of the other when

they perform tasks next to each other. The GROOP-effect experiment explores the

possibility that the action-perception relation in individuals is also influenced by

whether they feel themselves to be part of a group. Interpretation: in the congruent

condition, the participant and confederate see two left hands, which imply two

people working together, like them. In the incongruent condition, they see a left

and a right hand, which imply one person working alone. Evidence that the

participant represents the task as a contribution to something that the group

carries out, or as something to be performed by the individual alone, comes from

the effects of congruent and incongruent displays on the reaction time of the

participant. Reaction times are speeded in the congruent group condition

compared to the incongruent individual condition, even though, at the individual

level, the task of responding to the movements of the right hand on the screen is

identical. Reproduced, with permission, from [40].

interacting individuals to have the content of their indi-vidual actions specified by representing aspects of theinteractive scene in a distinct psychological attitude ofintending-together, believing-together, desiring-together,etc. [31]. This is the same as saying in the previous examplethat Mary and John understand what they do in terms ofMary doing her part and John doing his, only as part oftheir doing it together.

To articulate this insight, consider the capacity of indi-viduals to team-reason. Many episodes of sociality admit ofa rational explanation on the premise that agents makedecisions qua members of the group of which they arepoised to be a part (the ‘we’) [22,25,35–37]. For teamreasoning theorists, the claim that you see, or ‘frame’, youraction as part of an endeavour that we-as-a-team arepursuing together means that you are in the position towork out which course of action by all members of the groupis best suited to pursue the joint task [38]. You then actaccordingly by doing your own part [39]. This suggests thatby representing in the we-mode, my action as a member ofthe group will be guided by an individual-level represen-tation of what we are doing jointly (a ‘we-representation’).Indeed, according to the ‘GROOP effect’ [40], individualperformance varies in a joint task depending on whetherthe participants represent their task as a contribution tosomething that the group carries out or as something to beperformed by the individual alone (‘me-representation’)(Figure 1).

However, in order for a person to figure out her part as acontribution to something that is jointly achieved by allmembers of the group, agents must be able to co-representthe actions of their actual or potential interacting partnersby taking into account their perspective on the interactive

3 3

2 2

1 1

TRENDS in Cognitive Sciences

Figure 2. Co-representation and perspective-taking. Background: the figure shows

a situation designed to illustrate the role of perspective taking in developing the

co-representations necessary for cognizing in the we-mode. Interpretation: Mr.

Blue is sitting at a table with four mugs on it. For him, each mug on the table has an

affordance or action salience associated with it that indicates whether it is

potentially available for action [41]. Although the sharing of minds in the we-mode

requires more than just co-representing others’ viewpoints, co-representations are

automatically generated when there is a potential for joint action (Box 2). From the

point of view of Mr. Blue, mugs 1 and 2 have high affordance, because they are in

his reach, whereas mug 3 has low affordance, because it is out of reach. The

presence of a potential collaborator, Mr. Red, elicits co-representations by

inducing Mr. Blue to take into account the perspective of Mr. Red. The map is

changed to represent actions available to the ‘group’, so mug 3 now has a high

affordance, because it is in reach of Mr. Red [50]. By contrast, mug 2 now has a

lower affordance, since Mr. Red cannot see it [36].

163

Box 3. Questions for future research

� What are the differences, if any, between an account of the

processes that form and sustain shared intentions in terms of we-

mode processes and one that gives explanatory prominence to

the second-person perspective? What would the implications of

these differences be for psychiatric and pathological disorders of

social cognition?

� How can theoretical and experimental strands of research, such as

the team-reasoning and the co-representation literature, be

further integrated with the aim of generating empirically testable

hypotheses of the we-mode?

� Can a scientifically informed theory of the we-mode provide

conceptual and empirical insights into the nature and functioning

of social cognition as a whole?

� What are the methodological consequences of endorsing a theory

of the we-mode in social cognition for the design and implemen-

tation of experiments investigating the neural mechanisms of

social interaction?

Box 2. The ‘illusion’ of interaction

According to philosophical theories of joint action, individuals have

their minds shared when they interact intentionally [10]. On the

premise that the sharing of minds is formed and sustained by we-

mode processes, such as processes of co-representation, we argue

that cognizing in the we-mode captures the role of interaction for

social cognition, that is, the fact that agents have their potential for

social understanding and action expanded in the context of

interaction.

However, there is evidence that individuals tend to co-represent

the viewpoint of others in the interactive scene, even in situations

where they are just potential, rather than actual, interaction partners

[43]. Does this mean that cognizing in the we-mode is also possible

when agents are mere observers of others’ interactions? If so, what

would be the point of claiming that cognizing in the we-mode

explains how individuals have their social-cognitive resources

enhanced in the context of interaction, if one can be just a potential

interactant, namely, a spectator, and yet cognize in the we-mode?

To answer these questions, notice that being in a joint action is

sufficient for two agents to co-represent their perspective on the

action scene, but the converse does not hold. Hence, if an action is

to count as joint, there is more to the processes that underlie shared

mental states than just co-representation.

At least two agents must engage in the type of reasoning-as-a-we

(team) that underlies thinking and enacting things in the we-mode.

In more detail, individuals cognize in the we-mode when they see

their actions as contributing to something that they are doing

together and then act accordingly by contributing their own part. To

this end, the capacity to co-represent the others’ perspective is

essential for framing the scene of action. However, in addition to

this, mental states are shared in the case of joint action when at least

two agents engage in team-reasoning and explicitly reason that a

certain goal is best for the group and act appropriately on the

relevant intention [55].

Opinion Trends in Cognitive Sciences April 2013, Vol. 17, No. 4

scene (Figure 2). Co-representations serve various func-tions in joint action, such as providing control structuresand governing action monitoring during episodes of onlinecoordination [41,42]. Moreover, there is evidence that in-dividuals keep track of what the others do even in the merepresence of other people (Box 2), when interaction is notgoing to happen, and/or when co-representing others’ per-spective turns out to slow down an individual’s perfor-mance. This justifies the claims that priming a givensituation in the we-frame is not a matter of rational choiceand the we-mode might work as an implicit and automaticmechanism of mentalizing [43].

Interaction in the we-modeThe thrust of the interaction-based approach to socialcognition is that, when they are in the position to interact,individuals have their interpersonal understanding en-hanced through a ‘meeting’ of minds rather than an endlessascription of high-order mental states [7]. How, then,would a theory of shared agency in the we-mode explainthe fact that interacting individuals get to know moreabout each other’s propensities and disposition to act thanmere observers?

On our proposal, when action is performed by a group ofindividuals thinking in the we-mode, the social environ-ment adjusts agents’ potential for social cognition by pro-viding a broader understanding of the options available foraction, thus providing novel solutions for action. Co-repre-senting the others’ viewpoint on the action scene as acondition for acting jointly modulates the space of mental

164

activity and, therefore, behaviour, by providing each agentwith access to a set of descriptions and concepts that wouldbe unavailable from the observational, first-person singu-lar or third-person, perspective [44]. For example, actionsthat would not be available to me on my own are addedbecause they are available to someone else in my group[45–50]. Interestingly, this theory suggests that the mindis not just a product of the social: it is social all the waythrough. Human cognition is enriched with resources forcognizing in an irreducibly collective mode that remainlatent until individuals become engaged in particular in-teractive contexts. In this respect, the we-mode is a prop-erty of individuals but, since it manifests during activeparticipation in group behaviour, it cannot be understoodin purely individualistic terms.

We conclude by clarifying this claim in one importantway. This theory of social cognition is consistent withindividualism, because the we-mode is a mechanism orga-nized around cognitive and neural structures that areintrinsic to the individual and result from a dedicatedevolutionary and developmental history [32]. However,we do not assume that there is a ‘contrast’ between theindividual and the social, which can only be addressed bychoosing one side or the other. Rather, in line with mostrecent discussions in social cognitive neuroscience [8], wehave sought to integrate those levels into a mechanism-oriented approach to social cognition, while remainingcommitted to a non-reductionist view of collective psychol-ogy. Our suggestion is that social cognition is embedded inthe social environment to an extent that should be morecarefully pondered and theorized by individualistic-mind-ed scientists and philosophers alike (Box 3).

AcknowledgementsWe would like to thank four anonymous reviewers for their insightful andhelpful comments, as well as Francesco Guala, John Michael, and theInteracting Minds Group at the University of Aarhus, where earlierversions of the paper were presented and discussed.

References1 Frith, C.D. and Frith, U. (2012) Mechanisms of social cognition. Annu.

Rev. Psychol. 63, 287–3132 Apperly, I.A. (2010) Mindreaders: The Cognitive Basis of ‘Theory of

Mind’, Psychology Press

Opinion Trends in Cognitive Sciences April 2013, Vol. 17, No. 4

3 Hutto, D.D. (2004) The limits of spectatorial folk psychology. MindLang. 19, 548–573

4 Sebanz, N. (2006) Joint action: bodies and minds moving together.Trends Cogn. Sci. 10, 70–76

5 De Jaegher, H. et al. (2010) Can social interaction constitute socialcognition? Trends Cogn. Sci. 14, 441–447

6 Hutto, D.D. et al. (2011) Editorial: Social Cognition: Mindreading andAlternatives. Rev. Philos. Psychol. 2, 375–395

7 Butterfill, S.A. (2012) Interacting mindreaders. Philos. Stud. http://dx.doi.org/10.1007/s11098-012-9980-x

8 Schilbach, L. et al. Toward a second person neuroscience. Behav. BrainSci. (in press)

9 Michael, J. (2011) Interactionism and mindreading. Rev. Philos.Psychol. 2, 559–578

10 Roth, A.S. (2011) Shared agency, In The Stanford Encyclopedia ofPhilosophy (Spring 2011 edn) (Zalta, E.N., ed.), (http://plato.stanford.edu/archives/spr2011/entries/shared-agency/)

11 Searle, J.R. (1990) Collective intentions and actions. In Intentions inCommunication (Cohen, P.R. et al., eds), pp. 401–415, MIT Press

12 Tuomela, R. (2007) The Philosophy of Sociality: The Shared Point ofView, Oxford University Press

13 Hutto, D.D. et al., eds (2011) Social cognition: mindreading andalternatives, Rev. Philos. Psychol. 2, pp. 375–600

14 De Jaegher, H. and Di Paolo, E. (2007) Participatory sense-making: Anenactive approach to social cognition. Phenomenol. Cogn. Sci. 6, 485–507

15 Thompson, E. and Stapleton, M. (2009) Making sense of sense-making:reflections on enactive and extended mind theories. Topoi 28, 23–30

16 Grammont, F. et al., eds (2010) Naturalizing Intention in Action, MITPress

17 Gallotti, M. (2012) essay review on Naturalizing Intention in Action.Philos. Psychol. http://dx.doi.org/10.1080/09515089.2012.681868 (MITPress, 2010)

18 Menary, R. (ed.) (2010) The Extended Mind, MIT Press19 Herschbach, M. (2012) On the role of social interaction in social

cognition: A mechanistic alternative to enactivism. Phenomenol.Cogn. Sci. 11, 467–486

20 Gallagher, S. (2009) Two problems of intersubjectivity. J. Conscious.Stud. 16, 289–308

21 Singer, T. et al. (2003) Introduction: the study of social interactions. InThe Neuroscience of Social Interaction (Frith, C.D. and Wolpert, D.M.,eds), pp. xiii–xxvii, Oxford University Press

22 Bacharach, M. (2006) Beyond Individual Choice, Princeton UniversityPress

23 Wilson, D.S. et al. (2004) Cognitive cooperation – When the going getstough, think as a group. Hum. Nat. 15, 225–250

24 Colman, A.M. et al. (2008) Collective rationality in interactivedecisions: Evidence for team reasoning. Acta Psychol. 128, 387–397

25 Tomasello, M. (2009) Why We Cooperate, MIT Press26 Okasha, S. and Binmore, K., eds (2012) Evolution and Rationality:

Decisions, Co-operation and Strategic Behaviour, CambridgeUniversity Press

27 Postmes, T. and Jetten, J. (2006) Individuality and the Group.Advances in Social Identity, Sage

28 Hindriks, F. (2012) Team reasoning and group identification. Ration.Soc. 24, 198–220

29 Tomasello, M. (1995) Joint attention as social cognition. In JointAttention: Its Origins and Role in Development (Moore, C. andDunham, P.J., eds), pp. 103–130, Lawrence Erlbaum

30 Susswein, N. and Racine, T. (2008) Sharing mental states: causal anddefinitional issues in intersubjectivity. In The Shared Mind:

Perspectives on Intersubjectivity (Zlatev, J. et al., eds), pp. 141–162,John Benjamins

31 Gallotti, M. (2012) A naturalistic argument for the irreducibility ofcollective intentionality. Philos. Soc. Sci. 4, 3–30

32 Tomasello, M. (2008) Origins of Human Communication, MIT Press33 Butterfill, S. and Sebanz, N. (2011) Editorial: joint action: what is

shared? Rev. Philos. Psychol. 2, 137–14634 Tollefsen, D. and Dale, R. (2012) Naturalizing joint action: a process

based approach. Philos. Psychol. 25, 385–40735 Gold, N. and Sugden, R. (2007) Collective intentions and team agency.

J. Philos. 104, 109–13736 Samson, D. et al. (2010) Seeing it their way: evidence for rapid and

involuntary computation of what other people see. J. Exp. Psychol.Hum. Percept. Perform. 36, 1255–1266

37 Kovacs, A.M. et al. (2010) The social sense: susceptibility to others’beliefs in human infants and adults. Science 330, 1830–1834

38 Gold, N. (2012) Team reasoning, framing and cooperation. In Evolutionand Rationality: Decisions, Co-operation and Strategic Behaviour(Okasha, S. and Binmore, K., eds), pp. 185–212, CambridgeUniversity Press

39 Hakli, R. et al. (2010) Two kinds of we-reasoning. Econ. Philos. 26, 291–320

40 Tsai, J.C. et al. (2011) The GROOP effect: groups mimic group actions.Cognition 118, 135–140

41 Obhi, S.S. and Sebanz, N. (2011) Moving together: towardunderstanding the mechanisms of joint action. Exp. Brain Res. 211,329–336

42 Knoblich, G. et al. (2011) Psychological research on joint action: theoryand data. In The Psychology of Learning and Motivation (Ross, B.H.,ed.), pp. 59–101, Academic Press

43 Frith, C.D. (2012) The role of metacognition in human socialinteractions. Philos. Trans. R. Soc. Lond. B: Biol. Sci. 367, 2213–2223

44 Gallotti, M. Why not the first-person plural? Behav. Brain Sci. (inpress)

45 Newman-Norlund, S.E. et al. (2009) Recipient design in tacitcommunication. Cognition 111, 46–54

46 Bahrami, B. et al. (2010) Optimally interacting minds. Science 329,1081–1085

47 Isenhower, R.W. et al. (2010) Affording cooperation: embodiedconstraints, dynamics, and action-scaled invariance in joint lifting.Psychon. Bull. Rev. 17, 342–347

48 Griffiths, D. and Tipper, S.P. (2012) When far becomes near: sharedenvironments activate action simulation. Q. J. Exp. Psychol. 65, 1241–1249

49 Vesper, C. et al. (2013) Are you ready to jump? Predictive mechanismsin interpersonal coordination. J. Exp. Psychol. Hum. Percept. Perform.39, 48–61

50 Costantini, M. et al. (2012) Out of your hand’s reach, out of my eyes’reach. Q. J. Exp. Psychol. 65, 848–855

51 Ludwig, K. (2007) Foundations of social reality in collective intentionalbehavior. In Intentional Acts and Institutional Facts: Essays on JohnSearle’s Social Ontology (Tsohatzidis, S.L., ed.), pp. 49–71, Springer

52 Bardsley, N. (2007) On collective intentions: collective action ineconomics and philosophy. Synthese 157, 141–159

53 Tollefsen, D. (2004) Collective Intentionality, In Internet Encyclopediaof Philosophy, (http://www.iep.utm.edu/coll-int/)

54 Sebanz, N. et al. (2003) Representing others’ actions: just like one’sown? Cognition 88, 11–21

55 Pacherie, E. (2011) Framing joint action. Rev. Philos. Psychol. 2,173–192

165